The New Italian Criminal Law on AI. Deepfake, Copyright Infringement, Dangerous AI, and AI as an Aggravating Circumstance

The recently enacted Italian legislation on artificial intelligence, adopted in the wake of the EU AI Act, represents the first comprehensive national framework addressing the risks and opportunities of AI systems. Unlike the European regulation, the Italian law places significant emphasis on criminal law as a regulatory tool. It introduces new offences, such as the unlawful dissemination of AI-generated or altered content and deepfakes, and AI-related copyright infringements, as well as common and special aggravating circumstances when crimes are committed through AI. Moreover, the statute delegated to the government the task of redefining key principles of criminal liability, including the criteria for imputing responsibility to individuals and corporate entities, potentially reshaping the boundaries of both the general and special parts of criminal law. However, the Government has decided to exercise this delegated power only in part, and in the delegated decree approved, the general part of the Penal Code has remained untouched. While reaffirming the primacy of human decision-making, the legislation risks overburdening human agents with liability for harms effectively caused or facilitated by AI systems. This chapter critically examines these developments, highlighting the tensions between the Italian approach and the EU risk-based model, and discussing their implications for the future of criminal responsibility in the age of AI.